tryfonas, t. , & beale, d. (2018). exploration of the ... · management, agile methodology has...
TRANSCRIPT
Tryfonas, T., & Beale, D. (2018). Exploration of the Complex Ontology.Paper presented at 28th Annual INCOSE International Symposium,Washington DC, United States.
Peer reviewed version
Link to publication record in Explore Bristol ResearchPDF-document
University of Bristol - Explore Bristol ResearchGeneral rights
This document is made available in accordance with publisher policies. Please cite only the publishedversion using the reference above. Full terms of use are available: http://www.bristol.ac.uk/pure/user-guides/explore-bristol-research/ebr-terms/
Exploration of the Complex Ontology
Dean Beale MSc., C.Eng., C.Phys., M.IET.
Dept. of Engineering
University of Bristol
Bristol, UK [email protected]
Dr. Theo Tryfonas PhD
Dept. of Engineering
University of Bristol
Bristol, UK [email protected]
Copyright © 2018 by Dean Beale. Published and used by INCOSE with permission.
Abstract. The Oxford English Dictionary (OED), the established definition of words in the English
language, is at odds with other definitions of complexity proffered by Complexity Theory. This
variance is likely to cause confusion in the delivery community. The incorrect classification of a
project between ‘complicated’ and ‘complex’ is considered by some to be a major source of project
failure; resolving this issue is therefore critical. This paper explores the definition of complexity by
assessing definitions from various sources and by conducting a survey of over 100 delivery
professionals. The results demonstrate the extent of the confusion and have informed considerations
on how to resolve this. This paper recommends that it is essential that the definition is either defined
at the start, or that the term is avoided by using its components. This paper identifies a simple
modification to the OED definition that would resolve many of the issues, if implemented.
Introduction
Globalization and its associated information explosion mean that many delivery tasks are defined as
being complex. Consequently, delivery professions have risen to the challenge. The system
engineering profession prides itself on its ability to handle complex delivery tasks. In project
management, agile methodology has been developed, to a considerable extent, to accommodate
modern complex delivery requirements [1]. ‘Complex’ and ‘complexity’ have also become
buzzwords that justify significant further investments, an individual’s recognition, and the selection
of alternative approaches to delivery.
Despite this global-scale response to the rise of complexity, the term ‘complex’ itself is poorly
defined. This causes significant confusion. The term is often referred to as being difficult to define
[2]. Some have stated that one will know that something is complex when one sees it [3]. Meanwhile,
others stick to the dictionary definition, treating it at best as a synonym of ‘complicated’, which, for
many in the delivery community, is almost the antithesis of ‘complex’. A potential reason for this
confusion is that aspects of the Complexity Theory definition are increasingly becoming established
in the minds of the delivery community, but not to the full extent and detail that these theories define.
Consequently, it is possible that different aspects of these theories and definitions are established in
different minds, including potential over-simplifications.
The challenge of being able to define complexity with clarity is reflected in the complexity tools that
have emerged. Many use titles to define categories of complexity that are unique to the tool; e.g. cow,
bull, horse [4], or wicked, messes, and wicked messes [5, 6]. Use of such terms appears to be the
byproduct of an immature lexicon to describe the types of difficulty or complexity being observed, or
a justifiable attempt to avoid contentious definitions. Some tools have ‘complexity’ in their heading,
axis and as an output, with different definitions for complexity being implied at each level without
explanation, confusing the reader and breaking category boundaries. Many simple project
categorization tools include ‘simple’, ‘complicated’, ‘complex’ and ‘chaotic’ [7, 8, 9, 10] as the
categories. Although these tools demonstrate support for there being a different meaning for a
complicated and a complex system, supporting Complexity Theory, they run contrary to dictionary
definitions for complex, complicated and chaotic, potentially having a profound impact on
maintaining the consistency of commonly accepted terms.
The absence of an agreed ontology has not gone unnoticed, and has led some to determine that there is
no rigorous definition [11] or that the term is user-specific. This lack of clarity suggests either that a
clear definition is impossible or that the current definition may not be complete.
If the definition of these terms was inconsequential then the confusion might be more acceptable.
However, it has been noted by Michael Cavanagh of The International Centre for Complex Project
Management that the misclassification of a project as a ‘complicated’ task rather than a ‘complex’
one was a major cause of project failure [12]. A plethora of stories support the importance of
understanding the difficulty and adjusting to it. For example, it was a change of approach to
accommodate complexity that enabled both NASA to get the first man to the moon, despite trialing
for many years prior, and for the US Army to defeat AQI in Iraq [13]. Consequently, it would appear
that not only is the definition uncertain, but also that this uncertainty could be a major cause of failure
for many substantial projects across the globe.
The purpose of this paper is to examine the ontology of complexity by looking at the definitions as
used by a range of sources in the hope that a clearer definition or understanding can be developed.
This paper will focus on the objective assessment of delivery complexity with the aim of supporting
the delivery practitioner in identifying complex tasks effectively. The expectation is to move the
definition of delivery complexity discussion along, rather than to finalize it.
Relationship to existing theory and work
It is worth stating at the outset that the authors’ understanding of complexity is historically from a
delivery perspective. As part of the literature survey, we draw on insights from Chaos Theory and
Complexity Theory and the insights that can be gleaned from Difficulty or Complexity assessment
tools. The authors specifically avoid discussing the vast range of definitions on complexity that
emerge from Complexity Theory in the many professional bodies and delivery approaches, instead
aiming to use the salient points that have come to the fore that can provide insights into what
definitions are most recognized by delivery professionals and therefore can be used to help determine
whether a task is complex or not.
Approach
The approach taken in this paper is to consider definitions of complexity and associated words as
derived from the following sources:
1) Dictionary definitions (Oxford English Dictionary [14]; Collins Dictionary [15]).
2) Definitions as implied by developed complexity, difficulty or risk assessment tools that
explicitly deal with uncertainty. There are many project management complexity assessment
tools that use ‘complexity’ as a synonym of ‘complicated’; these have been ignored for the
purposes of defining complexity, as they provide no value.
3) Generic definitions as implied by common mathematical theories.
The definitions will be compared, contrasted and analyzed so that a suitable range of options can be
identified for clarifying the definition of complexity and/or its associated terms. The associated words
to be examined in addition to complex are: difficult, complicated, chaos, chaotic, emergent and
uncertainty.
From this analysis, a survey is constructed that tests the prevailing view of over 100 delivery
professionals from both the public and private sectors. The results are presented and conclusions and
recommendations are drawn from the analyses and results.
Literature review
In describing the outputs of the literature review it is not possible to describe one element without
using definitions from other elements. Consequently, it is not possible to order these definitions such
that the reader can move from one definition to the next with a full understanding. Instead, all
definitions need to be read and understood to fully understand each one. Some terms are well defined,
but a discussion of all of them is required to put the definition of complexity into the right context.
The tables below are RAG coded. The table cell color indicates the alignment of the definition within
that source; the alignment column indicates the alignment between the different sources of definition.
For example, a definition can be aligned within all three sources of definition, but those different
sources can be at odds with each other. Red indicates disagreement between the definitions; amber
indicates inferred differences; green means largely aligned.
Difficulty:
Figure 1. Table detailing the definitions of ‘difficulty’ from dictionary, tools and mathematical
theories. The RAG color indicates the amount of alignment in definition.
Source Definition Alignment
Dictionary: OED: Needing much effort or skill to accomplish, deal with, or
understand.
COLLINS US: Hard to do, make, manage, understand, etc.; involving
trouble or requiring extra effort, skill, or thought.
COLLINS UK: Not easy to do; requiring effort; a difficult job; not easy
to understand or solve; intricate; a difficult problem; hard to deal with;
troublesome.
Tools Aligned to above.
Theories Not discussed.
This term is explored because of its ability to replace the use of the ‘complex’ term in the title of many
tools. Often tools are called ‘complex’, suggesting that they measure the amount of complexity in a
task. However, their output is typically ‘simple’, ‘complicated’, ‘complex’ or ‘chaotic’. This suggests
that they indicate that the amount of complexity as you move from ‘simple’ to ‘complicated’, to
‘complex’, and then to ‘chaotic’ is increasing. This can lead to confusion. One way to resolve this is
to use ‘difficulty’ as a measure/title instead. Difficulty is the amount of skill and/or effort required to
complete an activity. Classifying tasks as ‘simple’, ‘complicated’, ‘complex’ or ‘chaotic’ infers the
types and amount of skill or effort required to deliver the task.
Uncertainty:
Figure 2. Table detailing the definitions of ‘uncertainty’ from dictionary, tools and mathematical
theories. The RAG color indicates the amount of alignment in definition.
Source Definition Alignment
Dictionary: OED: Not able to be relied on; not known or definite.
COLLINS US: Lack of certainty; doubt; the state or condition of being
uncertain; an uncertain matter, contingency, etc. Definition of Certain:
Fixed, settled, or determined; sure (to happen, etc.); inevitable; not to
be doubted; unquestionable; not failing; reliable; dependable; unerring;
without any doubt; assured; sure; positive.
COLLINS UK: Positive and confident about the truth of something;
convinced; definitely known; sure; bound; destined; decided or settled
upon; fixed; unfailing; reliable.
Tools Sometimes synonymous with unfamiliarity (not know), sometimes
synonymous with unpredictability (not able to rely upon). Rarely are
both aspects of uncertainty treated. Often applied to the inputs of
system development (requirements and solution). The uncertainty of
the system that delivers tends to be treated in isolation.
Theories Output uncertainty is closely aligned with emergent behaviour.
‘Emergent’, however, is the space of unknown unknowns, whereas
uncertainty covers known unknowns as well.
Uncertainty is inherently related to complexity and chaos. Consequently, this term is popular as an
axis in delivery complexity tools. Typically, it is the unfamiliarity in the requirements (don’t know
what) and/or with the solution (don’t know how) [7, 8, 9] that is measured, as shown in Figure 3
below.
Figure 3. A representation of a simplified Stacy matrix [9] indicating the association of complexity
with uncertainty in the What (requirements) and the How (in this instance, technology), which is
synonymous with solution.
These tools are used as an indicator of uncertainty in the outcomes categorized as simple,
complicated, complex or chaotic, inferring that ‘complicated’ has a measure of uncertainty within it,
and that ‘complex’ is more uncertain, and hence more difficult, than ‘complicated’.
The challenge with the tools approach is that defining ‘complicated’ as containing ‘some uncertainty’
does not objectively fit the description in the OED or in Complexity Theory. However, it does fit
subjectively with how tasks and projects are delivered, in that complicated approaches can be used to
handle some uncertainty, and complex approaches are used to handle more uncertainty. The
challenge with this simplified view is that ‘uncertainty’ can readily be separated into uncertainty with
the current state (or familiarity) and uncertainty with the future state (unpredictability). The
uncertainty measured above appears to be associated with the familiarity of the task at the start; it
does not take into account the uncertainty during execution between the component parts
(unpredictability) of the system that makes the system (also known as the machine that makes the
machine, or M3) [16]. Typically, the M3 system is unpredictable due to human decision-making.
Both lead to unpredictability as to what the final product of the system to be made is (also known as
the machine to be made, M2M) [16] and how much uncertainty is inherent in that system.
This suggests that multiple types of delivery uncertainty exist that also need to be considered. These
can be considered collectively or in isolation, as described in Figure 4 below.
Figure 4. A table to indicate how input familiarity and system unpredictability combine to create
increasing levels of uncertainty in a system output.
Known and predictable
delivery system
Unknown or unpredictable
delivery system
Familiar with how
and what
1. Deterministic,
predictable outcomes
2. Uncertain outcome
Familiar with how
or what
3. Uncertain outcome 4. Highly uncertain outcome
Unfamiliar with
how and what
5. Highly uncertain
outcome
6. Extremely uncertain
outcome
Emergent (Emergence):
Complexity Theory, on the other hand, tends to discuss ‘emergence’ instead of the unpredictable
aspect of ‘uncertainty’. This term is popular in systems engineering. However, in the delivery
community, some confusion could arise between ‘uncertainty’ and ‘emergence’, from the
philosophical definition of ‘emergence’ as used in Complexity Theory, and the more commonly
understood meaning, which is aligned to the Middle English or US definition (see Figure 5 below).
Figure 5. Table detailing the definitions of ‘emergent’ from dictionary, tools and mathematical
theories and other sources. The RAG color indicates the amount of alignment in definition.
Source Definition Alignment
Dictionary: OED: 1. In the process of coming into being or becoming prominent. 2.
Philosophy (of a property) arising as an effect of complex causes and
not analysable simply as the sum of their effects. 3. Middle English:
Occurring unexpectedly.
COLLINS US: Arising unexpectedly or as a new or improved
development; recently founded or newly independent.
COLLINS UK: Coming into being or notice; (of a nation) recently
independent.
Tools If discussed, more in terms of OED 3 or COLLINS US above (arising
unexpectedly).
Theories Emergence often discussed in Complexity and Chaos Theories as
defined in OED definition 2 above.
Other The whole is more than the sum of the parts, non-linear. [11]
Emergence in Complexity Theory differs from uncertainty in that it is focused around the unknown
unknown aspects of the outputs (predictability), whereas uncertainty covers the known unknowns and
the unknown unknowns of both familiarity and predictability. As such, it is either a subset of or a
synonym for ‘unpredictability’, depending on which OED definition is used.
Subjectively, however, an inability or unwillingness of the observer to analyze what the sum of the
effects of a system is means that it will often not be possible to separate the terms. As Complexity
Theory thinking, and hence the term ‘emergence’, permeates the thoughts of the delivery community
without proper introduction, there is a risk of confusion between the US/Middle English term
(unpredictability) and the philosophical term favored by Complexity Theory (unknown unknowns).
In addition, both the M3 and M2M system can exhibit philosophical emergence in addition to the
unpredictability that can be caused by the known unknowns.
Complicated:
The definition of complicated is universally agreed upon across dictionaries and in mathematical
theories as consisting of ‘many interconnecting parts or elements; intricate’. All definitions notably
exclude any reference to uncertainty.
Figure 6. Table detailing the definitions of ‘complicated’ from dictionary, tools and mathematical
theories and other sources. The RAG color indicates the amount of alignment in definition.
Source Definition Alignment
Dictionary: OED: Consisting of many interconnecting parts or elements; intricate.
Antonym = Easy, simple, straightforward.
COLLINS US: Made up of parts intricately involved; hard to untangle,
solve, understand, analyze, etc.
COLLINS UK: Made up of intricate parts or aspects that are difficult to
understand or analyze.
Tools Some tools infer [7, 8, 9, 10] that complicated systems have some
uncertainty. Others assume no uncertainty [16].
Theories Generally discussed as the absence of uncertainty.
In Complexity Theory, the definition of an intricate system without uncertainty is complicated. So
‘complicated’ is strictly defined as not having uncertainty. The dictionary makes no reference to
uncertainty. This is important because the dictionary description of ‘complex’ infers or states that it is
a synonym of ‘complicated’. As discussed above, some tools, such as those shown in Figure 1,
indicate that there is some uncertainty in complicatedness. It is assumed that these tools are using the
subjective delivery definition, which accommodates the fact that complicated delivery tools are able
to cope with some uncertainty, as demonstrated by the importance of risk logs. Indeed, it can be
argued that the greater the skill of the practitioner, the more uncertainty he or she can handle using
tools designed for a largely complicated task.
Chaos (Chaotic):
Figure 7. Table detailing the definitions of ‘chaos’ from dictionary, tools and mathematical theories
sources. The RAG color indicates the amount of alignment in definition.
Source Definition Alignment
Dictionary: OED: 1. Complete disorder and confusion. Antonym = Order. 2. The
property of a complex system whose behaviour is so unpredictable as to
appear random owing to great sensitivity to small changes in
conditions. Chaotic systems that exhibit either 1 or 2 above.
COLLINS US: 1. Extreme confusion or disorder. 2. Ancient
Mathematics: a pattern or state of order existing within apparent
disorder, as in the irregularities of a coastline or a snowflake.
COLLINS UK: Complete disorder; utter confusion.
Tools Significant uncertainty in the requirements and solution of a task that
combine to create chaotic outcomes. A combination of high
unpredictability, intricacy and unfamiliarity that means that the outputs
are unlikely to be aligned to expectations.
Theories
Chaos Theory: A system that appears random due to the high
sensitivity of the input parameters, although in actual fact the system is
deterministic and hence repeatable if the exact same inputs are used.
Otherwise it appears random.
Complexity Theory: Chaos is an extreme form of complexity; i.e.
highly emergent.
Chaos Theory definition requires absolute predictability in the system. As such, it falls outside the
Complexity Theory definition of a complex system, which mandates unpredictability or the
non-deterministic nature of the system. This Complexity Theory definition does include chaotic
systems that are non-deterministic. A chaotic system produces outputs that are so unpredictable, even
if repeated exactly, that they seem unrelated with the inputs. This is treated as a subset of a complex
system where the unpredictability or emergence is extreme. A Chaos Theory system, however, only
emulates this system, while it is in fact a deterministic system and hence is repeatable. Consequently,
the Chaos Theory definition does not match the dictionary definition of chaos as a subset of a
complex system, or as a system with complete disorder and confusion. However, the OED definition
of chaos uses terminology that indicates that it directly references Chaos Theory, albeit notably minus
the deterministic clause. Consequently, it appears that the definition of chaos in the OED responds to
the Complexity Theory definition of emergence, but actually uses unpredictability instead.
The prevalent use of ‘unpredictability’ suggests that a soft or adulterated form of Complexity Theory
definition is being established where ‘unpredictability’ replaces ‘emergence’, and where many of the
other aspects of Complexity Theory definition, such as context and history-specific and feedback
loops are simplistically folded into the ‘unpredictable’ banner. One could consider this a ‘soft’
Complexity Theory definition.
An example of this in the tools, as illustrated in Figure 3 above, is where chaos is defined as
significant uncertainty (unfamiliarity) in the requirement and solutions space [a], where a complex
system shows only some uncertainty (unfamiliarity) in these two elements. This indicates that a
chaotic system is an extreme form of a complex system with more uncertainty (unfamiliarity), and
that the definition of ‘chaotic’ as an extreme form of ‘complex’ aligns to all definitions. However,
chaotic and complex systems focus on the unpredictability or emergence in the system, not the
familiarity discussed in these tools.
Complex:
Figure 8. Table detailing the definitions of ‘complex’ from dictionary, tools and mathematical
theories and other sources. The RAG color indicates the amount of alignment in the definition.
Source Definition Alignment
Dictionary: OED: Consisting of many different and connected parts; not easy to
analyse or understand; complicated or intricate. Antonym = Simple or
straightforward.
COLLINS: Consisting of two or more related parts; not simple;
involved or complicated.
Synonym note: ‘complex’ refers to that which is made up of many
elaborately interrelated or interconnected parts, so that much study or
knowledge is needed to understand or operate it [a complex
mechanism]; ‘complicated’ is applied to that which is highly complex
and hence very difficult to analyze, solve, or understand [a complicated
problem]; ‘intricate’ specifically suggests a perplexingly elaborate
interweaving of parts that is difficult to follow [an intricate maze];
‘involved’, in this connection, is applied to situations, ideas, etc. whose
parts are thought of as intertwining in complicated, often disordered,
fashion [an involved argument]. The opposite of ‘complex’ is ‘simple’.
Tools
Some uncertainty in the requirements and solution of a task [7, 8, 9,
10]. Complex is a task with considerable uncertainty, which is however
possibly manageable using the right approaches by acceptance and
embracing of that uncertainty as in Agile delivery methods. Intricacy
and unfamiliarity in the task are sufficiently high that unexpected or
emergent outcomes may arise [db]. It is considered much more
challenging to handle than complicated systems. Cynefin definition
[17] reflects Complexity Theory definition.
Theories Difficulty to define, but inconclusively specified as a system that
exhibits:
1. Emergence (see above).
2. Is non-deterministic.
3. Has feedback and hence is able to resist or amplify change (is
self-healing as in rainforests).
4. Not necessarily complicated (intricate).
The opposite of complexity is clarity.
Other A complex system exhibits emergence [11]. The whole is different
from the sum of the parts, history matters, sensitive to context,
emergent, episodic (activity in fits and starts) [boulton].
Project managers characterize it as complicated + uncertainty [2],
adaptive systems, self-organization & emergence [2].
‘Complex’ as defined in dictionaries is essentially a synonym of ‘complicated’, with its opposite
being ‘simple’ (Collins). The dictionary definitions are closely aligned. However, the Complexity
Theory definition is also mature, although notably not finalized, and agreed across many
communities. These two definitions are at odds with each other. This is most obvious when looking at
the Synonym note in the Collins dictionary, which explicitly states that ‘complicated’ is a more
challenging form of ‘complex’; i.e. ‘complex’ is hard to understand and ‘complicated’ is very hard to
understand. This directly contradicts the delivery community’s understanding and tools usage of the
term.
Many delivery methods for handling complexity are aligned closely to the Complexity Theory
definition in that it has emergence or unpredictability as a key element. However, this alignment often
does not go down to the exact description of ‘complex’ as described by Complexity Theory. In
particular, tools and methods appear to use a soft form of ‘complex’ as specified by Complexity
Theory in that ‘emergent’ is synonymous with ‘unpredictability’ in the round.
Consequently, tools are roughly aligned to Complexity Theory, but Complexity Theory is in
complete disagreement with both dictionaries, particularly Collins. As the difference between these
definitions leads to different delivery methods, this is critical to resolve or clearly understand at least
from a delivery community perspective. It has already been mentioned that the misclassification of a
project as ‘complicated’ instead of ‘complex’ is considered by some the main source of project failure
[2]. The confusion caused by the use of an alternative definition throughout much of the delivery
community to that used in dictionaries can only exacerbate the issue.
Analysis
As can be seen, based on current definitions, it is not possible to resolve the definitions of complexity,
chaos and complicated systems without breaking one of the associated OED definitions, and/or
stepping out of line with the developed theories. Either these issues need to be resolved or the full
ambiguity of these terms needs to become more commonly understood and communicated for clear
discussions around complexity to take place.
By analyzing all the terms reviewed above it is hoped that one or more suitable solutions to resolving
the definition of complexity can be identified, around which the community might coalesce.
Summary of the issues
Before we start the analysis it would be valuable to summarize the issues.
1) Dictionary definitions are not aligned: ‘Chaos’ is defined as ‘a complex system whose
behaviour is so unpredictable as to appear random owing to great sensitivity to small changes in
conditions’. This suggests that a complex system typically exhibits unpredictable behaviour, and that
a chaotic system is an extreme case of this. The definition of ‘complex’ is synonymous with
‘complicated’, with no reference to unpredictability. Collins Dictionary goes a step further and
suggests that a ‘complex’ problem is easier to deliver than a ‘complicated’ problem. These definitions
seem to contradict one another.
This issue can also be considered by looking at the opposites. The definition of chaos indicates that a
complex system has unpredictability; hence the opposite would be ‘predictability’. The definition of
complex indicates that the system is intricate; the opposite is ‘simple’ or ‘straightforward’, as is the
case for ‘complicated’.
In addition, the definition of emergence, a form of uncertainty, indicates that it arises from a complex
system. It appears that Complexity Theory definitions have been identified in some terms within the
dictionary, but not in the complex or complexity terms.
2) Dictionary and Complexity Theory definitions of complex are not aligned: Complexity as
defined in the dictionary does not align to the Complexity Theory definition. The increasing
pre-eminence of Complexity Theory means that this clash should not be ignored. It is possible that the
emergence of Complexity Theory ideas among delivery community members who have not
otherwise studied it is causing the confusion. However, it appears that the soft form of ‘complex’ as
defined in Complexity Theory is emerging, in part because it is not possible to define ‘complex’ as in
Complexity Theory properly in less than a page or two, and that the definition is itself contended.
3) Chaos Theory is not a complex system: Complexity Theory states that a complex system is
emergent: the sum total of its parts cannot be used to predict its outcome; i.e., it is not deterministic. A
Chaos Theory system is specifically a deterministic system where the sum total of all its parts can be
used to predict its behaviour, but due to the hyper-sensitivity of the inputs it looks like a complex
system. It is explicitly a counterfeit complex system. This means that in the description of ‘chaos’ the
OED references a complex system, but largely uses the definition of a Chaos Theory system, minus
the term ‘deterministic’. The absence of this term means that one must assume a complex system even
though the terminology infers a Chaos Theory system. This first description of complete or extreme
disorder or confusion aligns well with Complexity Theory definitions as chaos as an extreme form of
complexity that has emergent (or unexpected outcomes).
Survey structure
To further analyze the definition of complexity as observed by the delivery community, a survey was
constructed based on the above discussion. The focus of the survey was to:
1) identify what definitions were most recognized by the professional delivery community and
consequently determine how best to communicate and discuss complexity and its associated
terms.
2) determine to what extent the Complexity Theory definition had permeated this community in
hard or soft form.
To achieve this, the dictionary definitions, along with definitions that reflected both the hard and soft
forms from Complexity Theory, and the tool definition inferred by Figure 1, were presented to over
400 delivery professionals in the public and private sectors, with over 100 responses split between
system engineers and project managers. The questions asked were:
Question 1) Please indicate in order of preference [1, 2, 3, etc] these definitions of system complexity
that you agree with. If you disagree, please indicate with a ‘d’.
a. Consisting of many different and connected parts, not easy to analyze or understand,
complicated, intricate.
b. Consisting of parts where the whole is different (greater or less) from what could be
determined by the sum of the parts, exhibiting feedback mechanisms, where the
outcome is also dependent on the context and history.
c. Consisting of many different and connected parts, not possible to fully analyze or
understand, leading to uncertainty in the outcome.
d. Consisting of any elaborately interrelated or interconnected parts, so that much study
or knowledge is needed to understand or operate it [a complex mechanism]; whereas
complicated is applied to that which is highly complex and hence very difficult to
analyze, solve, or understand [a complicated problem].
e. A system/task where some uncertainty in the requirements and the solution makes it
difficult to deliver, where more uncertainty in the requirements and the solution would
make it chaotic to deliver and less uncertainty would make it complicated.
f. Other: Please specify______________________________________________
Answer (a) is the OED definition of complexity. Answer (b) reflects the Complexity Theory
definition in a few words using key principles. As noted above, these definitions typically take many
paragraphs, so any attempt to condense them will be considered a poor imitation. The use of a fuller
definition was considered prohibitive to being able to conduct the survey; consequently, the aim was
to be close enough. The answer purposely does not use the term ‘emergent’; instead, the definition of
emergent was used. The reason for this, as discussed above, is that the definition of emergent is
ambiguous too; therefore we used the Complexity Theory definition of emergent to reduce confusion.
Answer (c) is an extended OED version and was designed to test the acceptance of a soft version of
complexity, as discussed above, with minimal change. Again how best to do this is not readily
obvious and is subject to interpretation; however, it only needed to be close enough to indicate the
intention. Answer (d) is a clarifying note in the Collins Dictionary. Answer (e) was designed to reflect
the diagrams used in delivery methodologies to determine whether a task is complex or not, as shown
in Figure 3. It is useful to see whether the Figure diagram, which is often presented and readily
accepted, was equally accepted when written down in text, forcing a more objective response.
Answer (f) was used to check that no obvious definition had been missed.
A second question was also asked.
Question 2: Please indicate the level of difficulty associated with the following words [1 = not
difficult; 4 = most difficult]: complex, chaotic, simple, complicated.
This question was asked to check the validity of the assumption that ‘complex’ is considered more
difficult than or equally difficult to ‘complicated’, a principle supported by all the definitions, as
illustrated in Figure 3, apart from the Collins note, which suggests that ‘complex’ is less difficult.
This question can also be used to check whether respondents had read the Collins definition correctly,
as it is possible for the answer to question 2 and Collins Dictionary to contradict each other. In
addition, the survey was introduced as a one-minute activity; however, the Collins note is considered
by the authors to be too complicated for a quick survey. Observation of the contradiction can be used
to indicate whether the respondents were using their intuition (or system 1) or their logical thinking
(system 2) [18] to respond to the survey.
Results
The results to question 1 of the survey are shown below in Figure 9. To assess the level of alignment
to each definition, the top two preferred definitions of each respondent were summed and compared
to the number of respondents who disagreed with the same definition.
Figure 9. A graph to indicate the number of respondents who selected each definition within the top
two preferred definitions of complexity and the number of respondents who indicated that they
disagreed with the same definition.
It can be seen from Figure 9 that the tools that are accepted subjectively are largely rejected when
assessed objectively. This does not mean they are not useful subjectively, however. The Collins note
is also highly controversial. The most relatable definitions, the OED and the extended OED, however,
still had more than 10% of the respondents directly disagreeing. This indicates confusion and a lack of
alignment of the definitions across the delivery community.
The results can also be analyzed via both community and sector, as shown in Figure 10 below.
OED ComplexityTheory
ExtendedOED
Collins Note Tools
53%
39%
57%
26%20%
11%15% 13%
24%29%
Responses from all communities
Top 2 preference Disagree
0%
10%
20%
30%
40%
50%
60%
70%
Public Sector responses
Top 2 preference Disagree
-10%
0%
10%
20%
30%
40%
50%
60%
70%
Private Sector responses
Top 2 preference Disagree
Figure 10. Graphs to indicate the number of respondents who selected each definition within the top
two preferred definitions of complexity and the number who indicated that they disagreed with the
definition, from the public sector, private sector, project management and systems engineering
communities.
Both private- and public-sector communities showed similar support for the OED definition. The
difference appears to be the acceptance of the Complexity Theory definition. The private sector
preferred the full Complexity Theory definition, while the public sector strongly preferred the
extended OED version. Comparing the systems engineering and project management communities,
the acceptance of the Complexity Theory definition is again the prevailing difference: the systems
engineering community supported it in first place, while the PM community ranked it in fourth place.
These results indicate strong community differences within the delivery community on terms that
they related to.
About 10% of the respondents provided alternative definitions. Many of these were alternative forms
of the Complexity Theory definition, such as the INCOSE or Cynefin [17] definitions. Some
provided added clarity to the extended OED definition with the addition of uncertainty with the inputs
or familiarity of the system. These responses, principally from system engineers, support the
hypothesis that producing effective definitions including Complexity Theory concepts is challenging.
A few genuinely new approaches to defining these terms were also proposed that were insightful, and
could be a better starting point for the definition of complexity. However, there is a concern that
increasing the number of competing definitions may cause more issues. The challenge is that, despite
many having strong views on what the definition is, these views are not typically the views of others.
In response to question 2, 38% provided responses that disagreed with ‘chaotic’ being more difficult
than ‘complex’, ‘complex’ being more difficult than ‘complicated’, and ‘complicated’ being more
difficult than ‘simple’. 17% of respondents explicitly indicated that a ‘complex’ task was easier to
deliver than a ‘complicated’ task, supporting the Collins note, but countering many of the definitions
of ‘complex’. This result is surprising and again underlines the importance of avoiding confusion.
44% of respondents’ answers to question 2 and question 1(e) conflicted, suggesting that the short
timeframe associated with the survey drove a system 1 intuitive assessment.
-10%
0%
10%
20%
30%
40%
50%
60%
70%
Systems Engineering responses
Top 2 preference Disagree
0%
10%
20%
30%
40%
50%
60%
70%
Project Management respnoses
Top 2 preference Disagree
Discussion
The results above indicate that there is significant opportunity for confusion around the definition of
key terms, especially across communities. It suggests that the misclassification of a project as
complicated rather than complex could readily be achieved due to misunderstanding the definition of
a complex system.
In order to communicate more effectively when we discuss complex systems we need to either: 1)
define what we mean each time with each audience; 2) avoid the term altogether, perhaps using
component parts such as intricacy, unfamiliarity and unpredictability [16] and how they contribute
towards making it difficult; or 3) align the definitions.
Option 1 above fails when the ‘local’ definition is consistently reinforced. Consequently, option 2 is
the most suitable approach short term. Option 3 is a longer-term approach with four options:
a) Keep the OED definition.
b) Support and wait for the Complexity Theory definition to establish itself.
c) Extend the OED version to accommodate uncertainty.
d) Propose a new definition.
Option a) is still largely the current state. This approach is being eroded by Complexity Theory
definitions; this needs to be reflected in the OED definitions.
Option b) defining complex as in Complexity Theory (hard) typically takes many paragraphs to
explain, and even then it is recognized as not fixed, complex and elusive. Consequently, a commonly
understood definition is likely to be evasive, even as the definition is established, unless it is
substantially simplified.
Option c) has significant benefits, although it does not solve all the problems. Adding uncertainty or
unpredictability to the OED definition supports the soft form of Complexity Theory definition and
would allow the hard form to co-exist with the modified OED version. It essentially unifies the space
with only a minor amendment. It resolves all three issues listed above, fixes the implied difference
between the OED definition of ‘chaos’ and ‘complex’, and allows Chaos Theory to be considered a
complex system, even though it is a unique case. The survey also suggests that the extended version is
the most acceptable definition to delivery professionals.
Option d) is appealing; however, this approach, without any globally authority establishing it, would
allow a swathe of competing strongly held definitions to propagate, exasperating the Complexity
Theory challenges in seeking consensus, and ensuring that the other person understands you further.
Conclusions and recommendations
It can be concluded from this work that, despite the importance of understanding what a complex
system is so the difficulty can be handled and mitigated effectively, the definition of complexity is
confused both in literature and in practice. It can be concluded that, in the short term, delivery
professionals should seek to avoid using the term as it can cause confusion. When selecting delivery
approach elements it should be achieved by assessing the system to identify those aspects that lead to
difficulty in delivering customer requirements and identifying techniques that mitigate those
difficulty aspects.
In the long term, a range of options has been considered. It is concluded that the only option that
appears to resolve the issues in the definition of complexity is to extend the dictionary definitions to
include aspects of uncertainty; for example, unfamiliarity and unpredictability in the system or its
inputs, leading to unpredictability in the outcome.
Further work
The definition of uncertainty, unpredictability, and emergence in the M2M, M3 inputs, system and
outputs should be explored further than this paper has been able to.
In addition, it has been identified as part of this work that Chaos Theory and Complexity Theory share
references to sensitivity and determinism or unpredictability. Examination of the
determinism-sensitivity space in which both these definitions sit could also prove beneficial.
Acknowledgements
The authors would like to acknowledge the many conversations with delivery professionals and staff
across public- and private-sector organizations, Bristol University and at international conferences
for helping with the exploration of the definitions of these terms. The authors would also like to thank
those who took time to fill out the questionnaire to support the survey results.
References
[1] Scrum alliance from: www.scrumalliance.org/why-scrum
[2] Hass, K., 2009, ‘Managing Complex Projects A New Model’, Management Concepts Inc.
[3] Know it when you see it quote. [need to complete the ref]
[4] Little, T., 2005, ‘Context-Adaptive Agility: Managing Complexity and Uncertainty’, xxx [need to
complete the ref]
[5] Grint, K. clinical leader, 2008, ‘Wicked Problems and Clumsy Solutions: The Role of
Leadership,’ British Association of Medical Manager (BAMM), vol. I, no. II.
[6] Wicked messes paper. [need to complete the ref]
[7] Obeng, E., 2012, ‘World after Midnight,’ TED talks.
[8] Turner, J., Cochrane, R.,1993, ‘Goals-and-methods Matrix: Coping with Projects with Ill-defined
Goals and/or Methods of Achieving Them,’ International Journal of Management, Vol. 11(2),
pp. 99–102.
[9] Stacy, R., 2012, Tools and Techniques of Leadership and Management, Routledge.
[10] Snowden, D.J., Cognitive Edge. [need to complete the ref]
[11] Holland, J., 2014, Complexity – A very short introduction, Oxford University Press.
[12] Cavanagh, M., 2013, ‘Project Complexity Assessment’, ICCPM, Kindle.
[13] McChrystal, General S. (retired), 2015, Team of Teams, Penguin, UK.
[14] Oxford University Press, 2004, The Dictionary of English.
[15] Collins dictionary reference. [need to complete the ref]
[16] Beale, D., Young, M., 2016, ‘Initial Thoughts on Measuring and Managing Complexity’,
European TEMS Symposium.
[17] Snowden, D.J., Boone, M.E., 2007, ‘A Leader’s Framework for Decision Making’, Harvard
Business Review.
[18] Kahneman, D., 2011, Thinking Fast and Slow, Penguin Books, UK.
Biography
Dean Beale has spent his career, principally within QinetiQ, delivering complex and difficult projects
in the defence and security sector serving a variety of Government customers. These roles have
included; Technical Manager, Project Manager and Systems Engineer. Dean has published over 20
papers in his fields of interest and was a Visiting Lecturer at the University of Birmingham on Project
Management. Dean is currently studying a Ph.D in ‘Assessing complexity and difficulty in
organizational change with associated mitigation strategies’ at the University of Bristol.